Independent-interactive Knowledge Learning for Multi-source Domain Adaptive Object Detection

doi:10.1016/j.inffus.2025.103642

Independent-interactive Knowledge Learning for Multi-source Domain Adaptive Object Detection

2025-08-21

0

PRE

AI

DOI：10.1016/j.inffus.2025.103642

Original

Original request for help

Save

Abstract

En 中文

• A novel multi-source domain adaptive object detection method is proposed. • Dual attention feature alignment is used to mitigate negative transfer. • Class relationship matching enhances the discriminability of features. • Experiments on four tasks validate the superiority of the proposed method. .

Keywords:

multi-source domain adaptation

object detection

dual attention feature alignment

class relationship matching

negative transfer mitigation

Independent-interactive Knowledge Learning for Multi-source Domain Adaptive Object Detection

Abstract

Journal

Organization